Intel's Process Roadmap to 2025: with 4nm, 3nm, 20A and 18A?!
by Dr. Ian Cutress on July 26, 2021 5:00 PM ESTCustomers Customers Customers
As a roadmap announcement today, the focus isn’t so much on the customers but on the technology. Because Intel is moving into a phase where it expects its IFS offerings to compete against the established players, it has to consider its disclosures with respect to both its internal use and any external interest, which is a new concept for the company – at least on this scale compared to its previous foundry efforts.
Intel CEO Pat Gelsinger, in the company’s Q3 financial call last week, was keen to point out that they already have a large hyperscaler customer signed up for their next generation packaging technology, however today there would appear to also be another customer in the mix. Now we assume that Intel’s Foundry Services is talking to 100s of chip companies, big and small, but it doesn’t take much to sign an NDA to start to talk – what will be interesting is when customers start making commitments to using Intel’s facilities, and if any of those are volume orders.
As part of the announcement today, Intel held a little bit back from us, saying that they are saving some of the details specifically for the event that is going on as we publish this piece. All we know is that our draft press release has a big yellow bar that says ‘[customer news]’ on it, right next to Intel’s 20A process node details.
For reference, Intel 20A is a 2024 technology using first generation Gate-All-Around transistors, marketed as RibbonFETs, as well as backside power delivery, marketed as PowerVias. At this time Intel expects to have second/third-generation EMIB available as well as fourth-generation Foveros Direct. So if a customer is already committing to Intel 20A, there’s going to be a lot of potential here.
When the announcement is made, we will update this news article.
To conclude, Intel maintains that these roadmaps will showcase a clear path to process performance leadership* by 2025. It’s a tall order, and the company has to execute better than it has in recent memory - but that’s kind of why the company has rehired a number of former Intel experts and fellows in research, product design, and execution.
*as measured by performance per watt at iso-power
Here's a secondary comparison chart (compared to the one on page one) with all three main foundry offerings listed in each of the main segments that Intel has discussed today.
Facebook
Twitter
RSS
326 Comments
View All Comments
watzupken - Wednesday, July 28, 2021 - link
I think by now we all know that the XXnm naming convention don't mean anything. It does somewhat give you an indication there is some progress in the node, but that's all. In my opinion, I don't really think Intel's current 10nm is in reality, better than TSMC's 7nm. I know the CPU architecture is also a major factor in the efficiency, but just looking at Intel's Tiger Lake H vs AMD's Ryzen 5000H, it is clear that the former consumes a lot more power to reach its performance target. And at the higher power limit, it is not able to decisively pull itself away from the Ryzen processor. Even Alder Lake is rumoured to launch with a PL2 value of 228W, despite using the improved 10nm SuperFin is quite a high number compared to a proper 16 cores Ryzen 5950X.nadim.kahwaji - Wednesday, July 28, 2021 - link
Excellent Article, that's what keep Anandtech so special, i hope the other topis (Especially Graphics ;) ) will Catch soon..SaolDan - Wednesday, July 28, 2021 - link
This move reminds me of Dewalt when they went from torque to UWO. They couldn't compete in torque so they made up their own unit of measurement so they didn't look bad. 20A makes me think of 20Amps.Spunjji - Thursday, July 29, 2021 - link
Dyson with "air watts", too 😅back2future - Wednesday, July 28, 2021 - link
How do consumers follow these advantages on real systems? TDP and energy efficiency, benchmarking tools from chip companies or reviews of integration into different systems or scores/$ and per years of duty?yankeeDDL - Wednesday, July 28, 2021 - link
4 new nodes in 4 years.Very aggressive.
If they can do it, they might actually regain the tech leadership. Wow, 20A tech in 2025 should bring more cores, significant more speed at less power than today's CPU (and GPUs).
Can't complain if that happens.
NickConrad - Wednesday, July 28, 2021 - link
What happened to the December 2019 roadmap that put a 3nm target on 2025? Is this not an admission that they’ll already miss that target by at least a whole generation?Spunjji - Thursday, July 29, 2021 - link
Looks like it!MDD1963 - Sunday, August 1, 2021 - link
Considering the interval between 14 nm and 10nm mainstream desktop CPUs, I'd recommend we look at one step at a time, lest the next shrink step take an additional 7 years...Oxford Guy - Wednesday, August 11, 2021 - link
I’m more concerned with how much power is going to be sacrificed to pursue the AI/spyware route.That’s Apple’s big thing in particular but it’s an industry-wide trend. Individuals are data, products corporations believe are purchased by them when individuals fork over cash.
We have already seen Apple roll out, for instance, a file system that’s abominably inefficient with hard disks and slower than the old one with flash. Then there are all those groovy black box piggybacked chips. I have read that the preposterously-named T2 is now to be embedded in the M2 CPU, like AMD’s PSP and whatever Intel’s spy chip is called.
A new page in CPU reviews will have to be devoted to how much power and die space is going to AI/spyware tech — like making it easier for the government:corporate complex to not only scan your files and take extremely extensive data about them — but move forward with thought crime like ‘preventative indefinite detention’ — a policy then president Obama laid out in a speech.
China is working to force employees to grin correctly in order for office equipment to work. Ugh.